Is the implication here that you need to charge and users will leave you once you do? If you can make a product that's significantly better, then you should be able to charge. The thing I'd note for affiliate marketing as a business model is that for it to generate significant revenue, you need to have a lot of traffic while other business models can generate that much faster (subscriptions) or make you money based off of that traffic (ads) instead of how many products are purchased.
Your note on affiliate marketing is what makes your first statement potentially unachievable. How does a consumer "know" that a product is significantly better to the point of "worth paying for"? There's always another free (potentially ad supported) affiliate marketer (or 5) around the corner. (Also considering the "worst" version of this
"product" is an unskippable ad").
Fair. The best solution we've seen is building the product in some way where it's somewhat defensible, either through data, features that bigger players won't build, etc. and then using a subscription based model if users are willing to pay for that and value the searches high enough or using an ads based model if you're optimizing for traffic rather than pure value on each search.
We've looked a bit into CR and Wirecutter, not that deeply into CR yet though. I definitely used Wirecutter for a bit of things in the past, and they have a high level of trust that we'll need to seek to replicate.